Chemical Sensor Drift Compensation via Reference Readings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Metal-oxide chemical sensors in portable electronic devices suffer from drift issues, leading to variations in sensor signals over time even under identical environmental conditions, affecting the accuracy of analyte detection.
Innovation Solution
A method involving reference readings taken at predetermined times to determine compensation values using prediction models, which are applied to operational readings to mitigate drift, employing techniques like linear regression and excluding deviating values to improve signal stability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference readings are taken at predetermined points in time to determine compensation values, then drift impact on sensor output is reduced, but device complexity increases due to additional processing steps
Solution Approach 1:
The patent applies preliminary action by taking reference readings at predetermined points in time before operational readings are needed. These reference readings are used to determine compensation values in advance through prediction models, which then correct subsequent operational readings. This proactive approach reduces drift impact without requiring complex real-time processing during critical measurement phases.
Solution Approach 2:
The patent uses copying by creating prediction models that replicate the sensor's drift behavior based on historical reference readings. These models serve as virtual copies of the drift pattern, allowing the system to predict and compensate for future drift without directly measuring it in real-time. The compensation values derived from these model copies enable accurate correction of operational readings.
2Measurement precision
If prediction models are determined using linear regression on reference values, then compensation accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies parameter changes by transforming the raw reference readings into compensation values through linear regression analysis. This mathematical transformation changes the parameters from time-dependent drift values to corrected compensation values that can be applied to operational readings. The linear regression model efficiently captures the drift trend by fitting a linear relationship, providing accurate compensation without excessive processing requirements.
3Reliability
If deviating reference values are excluded from prediction model determination, then model reliability is improved, but data loss increases
Solution Approach 1:
The patent applies the taking out principle by extracting and removing deviating reference values from the dataset used for prediction model determination. These outliers are identified as values that significantly deviate from the expected drift pattern and would negatively impact model reliability. By selectively removing only these problematic data points while retaining the majority of valid reference readings, the system maintains high model reliability without excessive data loss.
Data Source
AI summary
In a method for operating a portable electronic device comprising a chemical sensor, at least one reference reading is taken by the chemical sensor at least one predetermined point in time. At least one compensation value is determined based on at least one reference value wherein each reference value is derived from a result of a corresponding reference reading. A result of an operational reading of the chemical sensor is modified by the at least one compensation value.